How AI Can Leverage EHR for More Efficiency at the Bedside
By Matt Phillion
A new solution from Lucem Health uses clinical AI to help identify patients who may be at a higher risk of serious or chronic illness. It does so by using readily available clinical data from the electronic health record (EHR) to surface insights that help providers and organizations improve patient outcomes.
Sixty percent of Americans live with at least one serious or chronic condition, such as cancer, diabetes, stroke, or heart disease. By expanding the opportunity for earlier diagnoses and offering more personalized interventions, providers have the chance to improve outcomes and quality of life for patients before a condition worsens and becomes more difficult to manage.
Clinical data is often an untapped resource for identifying patient risks hiding in plain sight, notes Sean Cassidy, founding CEO of Lucem Health. So how can leveraging tools like Lucem Health’s Reveal solution impact more patient lives?
The initial idea was a result of a collaboration with Mayo Clinic, Cassidy notes. “As you would expect, Mayo has done a lot of research into AI and machine learning,” he says. “The challenge for many providers trying to develop AI is, how do they get something from the bench to the bedside? How does it go from a research project into something operational, so it’s integrated in the workflow and helps clinicians deliver better care?”
Current platforms out there focus on different specialties, Cassidy notes, but a general-purpose, clinically focused AI deployment platform didn’t exist—which was the core idea behind what has become Lucem Health. But calling the Reveal platform AI, Cassidy says, doesn’t quite explain what it does and how it works.
“We almost do ourselves a disservice calling it AI,” he says. “There’s an implication that there’s something mysterious or anthropomorphic about AI. It’s true that these algorithms can be opaque, but at the end of the day it’s about pattern recognition, recognizing subtle trends in data that previously would have been hard to find. The way I think about it is that we are taking the statistical analysis capabilities we’ve had in the past and we’re supercharging them.”
The AI (or machine learning) is driven by having access to large quantities of patient data. “We’re very early in the cycle, but what we’re trying to do is create practical and pragmatic AI solutions that help physicians focus their attention on the patients who may be at higher risk for serious disease,” says Cassidy. The goal isn’t to take attention away from other patients, of course, but rather to flag patients who may have a condition or set of circumstances that puts them at higher risk.
Unobtrusive but effective
The idea behind Reveal is that it works quietly without distracting the clinicians it is assisting. “The typical clinician doesn’t even notice it working in the background,” says Cassidy. But with Reveal’s insight gained by analyzing existing data, they’ll know that when a particular patient arrives for a visit, “this patient may have a higher risk of diabetes, or if they already have prediabetes, that they’re at a higher risk of progressing to Type 2.”
The tool helps clinicians focus on delivering better, more proactive, higher-yield care, Cassidy explains.
“These are not robots taking over clinicians’ jobs. We believe that, because [clinicians are] able to focus on patients who may be at higher risk, they can practice medicine more at the top of their licenses,” says Cassidy. “In effect, we are increasing the likelihood that the patients they see in their exam rooms have a serious condition.” The idea is not to give practitioners more work or additional tasks, but rather to bring the patients in and let the doctors be doctors, he explains.
“On the administrative side, for those focused on population health and managing risk, we’ve heard that it’s the most practical use and deployment of AI or machine learning they’ve seen,” says Cassidy. “It integrates with existing workflows rather than disrupting them. We know clinicians are overwhelmed with alerts from the EMR, and we don’t want more of that, but when you’re able to identify which patients are more likely to have an issue, it’s enormously helpful for clinicians.”
Carmakers, not engine makers
One of the challenges to communicating the details of this technology is clarifying the distinction between what Reveal does and what the company’s partners do.
“The best metaphor I’ve come up with is that we’re carmakers, not engine makers,” says Cassidy. “We work with algorithm development partners who are data scientists with a tremendous amount of expertise. They make the engines. Our team doesn’t have data scientists. We work with those experts and innovators who, to continue the metaphor, need a partner who knows how to deliver the fuel to the engines and the engines’ output to the front lines of healthcare. That fuel is data, and the destination for the car is better care.”
Healthcare organizations, therefore, work to steer the car. “Mayo and others provide guidance on their priorities: chronic kidney disease, opioid use, COVID, etc. This will differ between provider organizations depending on where they sit geographically, what their payer mix is, what their population looks like,” says Cassidy.
Each organization’s unique needs help Lucem Health and its algorithm development partners create appropriate solutions. “We’re focusing on two areas. One is helping be a part of the process to detect diseases earlier,” says Cassidy. “The other is helping to optimize care delivery. We’re not a diagnostic tool; we’re helping providers improve the clinical and financial yield of care delivery processes they are already delivering. We help them get their patients into care pathways and on treatment plans that improve quality of life, which pays real dividends downstream.”
Many provider organizations are now investing in data science, machine learning, and AI, but are struggling to figure out how to scale it. “We work with providers who have a desire to take their own AI innovation from the bench to the bedside,” says Cassidy. The implementation details can vary greatly depending on the sophistication and needs of the partner organization.
“Providers are recognizing AI’s potential for them to get more value of their significant and ever-growing data assets,” says Cassidy. “We’re finding there’s a high degree of AI receptivity and literacy not only in healthcare IT departments but also with CMOs, CMIOs, and clinical department leaders.”
There’s a process to getting the most out of this technology, though. “First, ensuring clinicians see these tools as benign forces in their universe. Getting them to trust and adopt these tools, become more literate about them, and see them as indispensable to practicing at the top of their license,” says Cassidy.
Such tools can also help alleviate the staffing shortage, one of the ever-present issues in healthcare right now. “Clinicians are overloaded, they have high levels of stress, and COVID created more strain,” says Cassidy. “We want to help them be more productive, enjoy their jobs more, and better do what they signed up to do, which is taking care of people in their communities. I believe these tools have the potential to help them do all those things.”
Matt Phillion is a freelance writer covering healthcare, cybersecurity, and more. He can be reached at matthew.phillion@gmail.com.